Team:Leiden/Results
<!DOCTYPE html>
Rapidemic
Results
Once the mechanism of our detection technique was designed, we performed laboratory experiments to determine the experimental conditions of the new technology. After a few rounds of optimization, a proof-of-concept experiment showed how the reactions of the technology can be successfully connected to detect, amplify and report the presence of a target DNA. On this page, we show the experiments that led to the proof-of-concept in detail. The results were also published as a pre-print on BioRxiv; the complete publication can be viewed after the Discussion section.
Introduction
Recently, a novel severe acute respiratory syndrome (SARS), coronavirus disease 2019 (COVID-19), spread worldwide, causing over 27,800,000 cases of infection and 900,000 deaths as of September 10th, 20201. Although this was the most recent pandemic, it was not the first outbreak that caused major health issues on a global scale, and it will also not be the last one. Potential future epidemic-causing pathogens are listed in the WHO R&D Blueprint and also include unknown diseases, collectively named “Disease X”2. To be prepared for "Disease X" (and other future infectious diseases), we developed an innovative detection technique, called Rapidemic, that can be rapidly adaptable to new pathogens. This technique is based on an isothermal amplification method coupled to a colorimetric readout for nucleic acid detection. Due to its ability to be used in point-of-care settings, it has the potential to enable population testing worldwide.
The assay is built on three reactions to yield a color change visible to the naked eye (Fig. 1). The working mechanisms of these reactions are explained in detail on the Engineering page. In short, the first reaction is recombinase polymerase amplification (RPA). RPA is an amplification reaction that works at a low and constant temperature, during which a specific target sequence is exponentially amplified. The product of this reaction serves as a template for the second reaction of the assay, the linear strand-displacement amplification (LSDA) reaction. Here, a short single-stranded DNA sequence, a guanine-quadruplex (GQ) DNAzyme, is produced. During the last reaction of the assay, the DNAzymes catalyze an oxidation reaction to produce the color change. To validate the new detection technique, we first characterized each of the reactions individually. This was followed by a proof-of-concept study during which we coupled the three reactions.
Fig. 1 Amplification scheme of the DNAzyme-based label-free detection method Rapidemic. The Rapidemic detection method relies on three main reactions: RPA, LSDA, and GQ-hemin DNAzyme-catalyzed TMB oxidation. During RPA, the target sequence is amplified and extended with a short segment of nucleotides due to primer overhang. This segment contains the reverse complementary sequences of a recognition site for a nicking endonuclease (nickase) and a GQ sequence. The nickase makes a single-stranded cut near its recognition site in the amplification product, after which a strand-displacement polymerase elongates the 3’ end at the nickase cut site and thereby displaces the single-stranded GQ sequence, so that it is released from the double-stranded DNA molecule. The single-stranded GQ sequence forms a three-dimensional structure with peroxidase activity upon binding to potassium ions and hemin. Finally, the DNAzyme catalyzes the oxidation of 3,3′,5,5′-tetramethylbenzidine sulphate (TMB) in the presence of hydrogen peroxide (H2O2) to produce a color change. As a proof-of-concept for this principle, we reported the initial validation of this detection method with genomic DNA of S. cerevisiae. Figure adapted from our preprint.
Individual characterization of each reaction
First reaction: RPA
The RPA characterization experiments were aimed to:
+ Confirm amplification with designed primers
+ Study RPA kinetics at different temperatures
+ Scale down RPA reaction volume
? Determine RPA detection limit
? Resolve false-positive primer-associated amplification
(+ and ? indicate "successful" or "remains a challenge")
For most RPA experiments, we used isolated genomic DNA of Saccharomyces cerevisiae (S. cerevisiae) as the amplification target. We performed RPA using the commercial TwistAmp Basic RPA kit (TwistDx). To target a specific sequence on the genome of S. cerevisiae, we designed a set of primers using a freely available software program called PrimedRPA3. The primer sequences can be found in Table 3 on the Parts page. To couple the RPA reaction to the second reaction in our detection scheme (LSDA), we inserted a 27-bp overhang sequence at the 5’-end of the reverse primer. This overhang sequence contained the reverse complementary sequence of a nickase recognition site and the EAD2+3’A DNAzyme, separated by a short spacer ( <10bp).
The main results of our RPA experiments were:
- The RPA primer pairs were able to successfully amplify their target sequence (Fig. 2).
- The presence of 5’-overhang (reverse primer) did not interfere with the amplification of the target sequence (Fig. 2).
- The lowest temperature limit for RPA was 20 °C (Fig. 3).
- For > 10 minutes incubation time, the amplification yield does not depend on the incubation temperature in the range of (at least) 30 - 42 °C. For incubation time < 10 minutes, there is a tradeoff between high yield and low temperature (Fig. 4).
- The final RPA reaction volume is 20 µL.
At first, when we performed RPA reactions with a very low concentration of target DNA to test the sensitivity of RPA, the gel electrophoresis and real-time RPA results showed false-positive signals. The off-target amplification presumably caused the inconsistent results that we observed in these experiments (Fig. S2), and they were presumably caused by primer-associated amplification initiated through primer secondary structures and/or primer-primer interactions (Fig. 10b). The addition of 2.5% DMSO as an amplification enhancer to the RPA reaction successfully reduced the false-positive signals for S. cerevisiae, as analyzed by real-time RPA using the SYBR Green dye (Fig. 10c).
We then performed RPA reactions with small synthetic templates derived from various other microorganisms to show the versatility of our detection system. These targets organisms were S. cerevisiae, Bacillus subtilis, Plasmodium falciparum, Mycobacterium sp. and influenza A H1N1. However, only for S. cerevisiae and influenza A, amplifications were obtained without a false-positive signal when analyzed by real-time RPA using the SYBR Green dye (Fig. S3).
Second reaction: LSDA
The LSDA characterization experiments were aimed to perform the following:
+ Compare different nickases and identify the most efficient one
+ Reduce the nickase concentration and the LSDA reaction volume
? Study the reaction kinetics at different temperatures
The LSDA reaction is the bridge linking the first amplification step (RPA) to the signal reporting step (TMB oxidation). LSDA uses a nickase to produce specific single-stranded DNA sequences, namely GQ DNAzymes. For this, the nickase requires a template DNA molecule, which is the product of the RPA reaction. During the LSDA characterization experiments, we aimed to identify the performance of different nickases (Nt.BstNBI, Nt.AlwI, and Nt.BsmAI), for which we used synthetic templates (see Table 2 on the Parts page). Since the nickases are one of the most expensive components of our detection method (Fig. 3 on Proof of Concept page), we also aimed to optimize the nickase concentration and LSDA reaction volume. For that, we used self-generated RPA products.
The main results from these experiments were:
- The Nt.BstNBI was the most efficient nicking endonuclease (Fig. 5).
- The use of Nt.BstNBI was decreased 40-fold by reducing the nickase concentration 8-fold (Fig. S4) and the LSDA reaction volume 5-fold (Fig. S5).
We did not manage to study the LSDA kinetic rates at different temperatures due to a lack of time. We planned to use the kinetic parameters in our (semi-)empirical model that simulates the underlying reaction kinetics of our detection method. Instead, we adopted kinetic parameters from literature to build the model.
Third reaction: DNAzyme-catalyzed oxidation
The DNAzyme-catalyzed oxidation experiments were aimed to perform the following:
+ Confirm functioning DNAzyme (New Part added to iGEM Registry and Existing Part improved)
+ Determine kinetic parameters of DNAzyme to incorporate into model
? DNAzyme concentration calibration curve with optimized reaction conditions
The last reaction of our detection technique is the GQ DNAzyme-catalyzed TMB oxidation reaction. GQ-hemin complexes catalyze the oxidation reaction of 3, 3’, 5, 5’-tetramethylbenzidine (TMB) in the presence of hydrogen peroxide, resulting in a color change. As the color change is observable by the naked eye, we can enable the detection of pathogenic genetic material in a sample without the requirements for complex laboratory equipment. We tested a variety of GQ DNAzymes to select the most efficient one. We also varied the concentration of the substrate (TMB) to obtain the kinetic parameters of the most efficient DNAzyme.
The main results from these experiments were:
- The most potent GQ DNAzyme was EAD2+3’A (Fig. 6).
- The addition of an adenine base at the 3'-end of DNAzyme BBa_K1614007 greatly boosted its activity (Fig. 6).
- The molar maximum TMB conversion rate (kcat) of EAD2+3'A was 3.971x104 min-1 M(DNAzyme)-1, and the TMB affinity of EAD2+3’A was 6.798x10-3 g L-1 (Fig. 7 and Table 1).
Click here to go to the Results section of the individual characterization of the three reactions.
Proof-of-concept: coupling the reactions
The proof-of-concept experiments were aimed to perform the following:
+ Couple RPA, LSDA and GQ-catalyzed oxidation in a serial manner
? Combine RPA and LSDA in a one-pot reaction
? Determine the detection limit of the detection method
? Optimize reactions for low false-positive and false-negative rates
Having optimized each reaction separately, we aimed to couple RPA, LSDA and TMB oxidation reactions and validate our assay. Therefore, we passed the product of each preceding reaction to the next in a serial manner. Initially, we did not succeed in coupling all three reactions; no positive colorimetric signal from GQ-hemin catalyzed TMB oxidation was detected. Therefore, we started to investigate possibilities for optimization of the coupling step.
Based on the optimization steps we performed we could conclude the following:
- Purification of the RPA product prior to the LSDA reaction resulted in a colorimetric signal (Fig. S7); a compound in the RPA mix interfered with the oxidation reaction.
- The reducing agent DTT present in the RPA mix inhibited the oxidation reaction (Fig. S8).
- Lowering the pH of the oxidation buffer decreased the inhibiting effect of DTT on the oxidation (Fig. 8).
- Lowering the pH of the oxidation buffer was not sufficient to enable color generation; the amount of RPA product in the oxidation reaction was too high (Fig. 9a and b).
- Optimization of the dilution rates of the RPA and LSDA products enabled color generation (Fig. 9c and d).
- False-positive readouts were generated as a consequence of primer self-amplification and/or primer-primer interactions (Fig. 10a and b).
- The addition of 2.5% DMSO to the RPA mix showed to be an efficient strategy to reduce false-positive signal (Fig. 10c and d).
Click here to go to the Results section of the coupling of the three reactions.
Future Prospects
Text adapted from our preprint
Here we presented a novel design of a label-free DNAzyme-based detection method called Rapidemic. This assay combined RPA with linear strand-displacement amplification (LSDA) and guanine-quadruplex (GQ) DNAzyme-catalyzed color-changing reaction. The colorimetry basis of the signal readout omitted the need for extensive instrumentation. Moreover, primer-based sequence detection of RPA gave Rapidemic a potential to be rapidly adapted to target a new sequence. As a proof of concept, we developed the assay to detect isolated genomic DNA of Saccharomyces cerevisiae. The use of low-pH buffers and the optimization of the dilution rates from each preceding reaction to the next showed to be successful strategies to enable visible detection with this method. These findings demonstrate for the first time that a label-free DNAzyme-based detection method can be coupled to RPA and LSDA for nucleic acid detection.
Despite successful integration of the three reactions, the technique presented significant false-positive signals in the absence of target sequence. Although the addition of DMSO in the RPA reaction effectively reduced the occurrence of false-positive signals, the less distinct color change due to DMSO revealed the need to further improve and optimize the presented sequence detection method to be used as a rapid diagnostic tool. Future work may focus on alternative enhancing chemicals that do not interfere with the oxidation reaction, as well as improved primer sequences that are less prone to have off-target interactions.
Despite the need for further optimization, the presented sequence detection method holds potential to be a rapid diagnostic tool. Its RPA-based detection allows it to be repurposed to target other DNA sequence by simply changing its primers, similar to the development of PCR-based tests. This allows rapid adaptation of the method to target newly emerging diseases. In comparison, development of antibody and/or antigen-based tests may require significantly more time and resources to be redeveloped to target a new disease4. Additionally, the reagent costs of the presented detection method are similar to commercial PCR-based tests. However, unlike PCR-based tests, our method does not require extensive instrumentation to give a signal readout. Altogether, the method presents a potential to be a rapidly adaptable point-of-care diagnostic tool.
Future research may focus on combining RPA and LSDA into a one-pot reaction. The development of a one-pot reaction may further hasten the reactions’ incubation period and reduce the reagent costs (dNTPs, buffers, polymerase). Moreover, this would greatly simplify the hardware design when applied in point-of-care diagnostics. It may be necessary to replace the commercial TwistDx RPA kit with a custom-made RPA mix to optimize its composition for one-pot RPA and LSDA. With such custom-made mix, the concentration of DTT could be reduced to increase the amount of RPA product allowed in the oxidation reaction and increase the method’s sensitivity. Altogether, the presented sequence detection scheme presented a promising potency as rapid point-of-care diagnostic tool that may contribute in future attempts to diagnose, and by doing so, monitor and contain the spread of an infectious disease.
Click here to read the full discussion.
Visit our Proof of concept to see how we integrated our successful detection method into a preliminary prototype device.
References
- John Hopkins University and Medicine. COVID-19 map. John Hopkins Coronavirus Resour. Cent. (2020). Available at: https://coronavirus.jhu.edu/map.html. (Accessed: 10th September 2020).
- World Health Organization. WHO R&D Blueprint: list of blueprint priority diseases. (2020). Available at: http://www.who.int/blueprint/priority-diseases/en.(Accessed: 11th September 2020).
- Higgins, M., Ravenhall, M., Ward, D., Phelan, J., Ibrahim, A., Forrest, M. S., Clark, T. G., and Campino, S. (2019). PrimedRPA: primer design for recombinase polymerase amplification assays. Bioinformatics, 35(4), 682–684. https://doi.org/10.1093/bioinformatics/bty701.
- Peeling, R. W., Murtagh, M. & Olliaro, P. L. Epidemic preparedness: why is there a need to accelerate the development of diagnostics? The Lancet Infectious Diseases 19, 172–178 (2019).
RPA
Confirmation of target amplification using recombinase polymerase amplification (RPA)
Text and figure adapted from our preprint
Our method employed an initial target amplification step. This step is essential to selectively amplify the positive signal from the targeted sequence(s) present in the sample. Initial target amplification was performed using the TwistAmp Basic RPA kit (TwistDx). The primer pairs used in this experiment were designed to target a 191 bp sequence from the genome of Saccharomyces cerevisiae (S. cerevisiae). In addition to the target-specific sequences, the reverse primer contained a 27 bp overhang sequence at its 5’-end. This primer overhang encoded for a nickase recognition site, a short spacer (less than 10 bp), and a sequence reverse complementary to EAD2+3’A DNAzyme. Here, we evaluated the primer pairs’ ability to detect the presence of its target sequence by using S. cerevisiae BY4741 genome as target sample (Table S2). As can be seen from Fig. 2, the inclusion of 5’-overhang sequence did not appear to have interfered with RPA’s ability to detect the presence of its target sequence.
Primer sequences can be found in Table 3 on the Parts page.
Fig. 2 RPA with forward primer SC1_F_p1 and reverse primers SC1_R_Nt.AlwI_EAD2+3A (lane 2 and 3) or SC1_R_Nt.BstNBI_EAD2+3A (lane 4 and 5) targeting S. cerevisiae genome. Each reaction was performed using genomic S. cerevisiae BY4741 DNA as a target sequence. The target sequence is absent in the negative control. The positive control contains the primers and target templates provided by the TwistAmp Basic kit (TwistDx).
One of the two samples with reverse primer with the Nt.BstNBI recognition site did not work in the previous experiment (Fig. 2). This may have been caused by an experimental error. Therefore, we repeated the experiment, including additional samples with reverse primers with the recognition site for Nt.BsmAI or Nt.BspQI. This time, bands were observed at the expected height for all replicates for the Nt.AlwI and Nt.BstNBI primer sets (Fig. S1). For the Nt.BsmAI primer set, no amplification was observed. For the Nt.BspQI primer set, only one of the two replicates showed a band at the expected height. We continued our experiments with the Nt.BstNBI and Nt.AlwI primer sets.
Fig. S1 RPA with forward primer (SC1_F_p1) and reverse primers (SC1_R_Nt.AlwI_EAD2+3A or SC1_R_Nt.BstNBI_EAD2+3A or SC1_R_Nt.BsmAI_EAD2+3A or SC1_R_Nt.BspQI_EAD2+3A) targeting S. cerevisiae BY4741 genome. The reactions were performed in duplicate.
RPA kinetics at different temperatures
The recommended temperature for RPA reactions performed with the TwistAmp Basic RPA kit (TwistDx) is 37 to 42 °C1. However, performance at lower temperatures may be advantageous when the reaction is integrated for point-of-care use. Previous studies showed that RPA can work while it is heated by holding it in the palm of the hand2. Performing RPA at room or body temperature will significantly simplify the design of a diagnostic kit as the incorporation of a heating instrument would no longer be required.
As both incubation time and temperature affect the performance of RPA, several experiments were performed to test both variables in tandem. RPA reactions were performed at five different temperatures (4, 20, 25, 30 and 42 °C). Four identical samples were incubated in a PCR machine at each temperature. At intervals of 5 minutes, one sample was taken out of the PCR machine and placed on ice.
Fig. 3 confirms the Amplification of the expected target sequence was confirmed for all tested temperatures (Fig. 3). The samples incubated at 4 °C served as negative controls, and as expected no band was observed. For 20 °C, a minimum incubation period of 15 minutes was needed to confirm amplification of the target on the gel. However, for 42 °C, 5 minutes of incubation was enough to confirm amplification. When we compare the samples incubated for 20 minutes at different temperatures, we can observe that the highest yield is at 30 ºC. Thus, this experiment showed that RPA can work at temperatures as low as 20 °C, although the yield at 20 °C was significantly lower than for 25 °C to 42 °C. Further experiments are needed to show whether the amplification yield at 20 °C is sufficient for sensitive detection with our detection method (RPA coupled to LSDA and DNAzyme-catalyzed oxidation).
Fig. 3 RPA kinetics at different incubation temperatures. Forward primer (SC1_F_p1) and reverse primer (SC1_R_Nt.AlwI_EAD2+3A) targeted S. cerevisiae BY4741 genome. Positive and negative controls were incubated for 20 minutes. Amplification was not confirmed for the positive control at 25 °C, which could mean that this specific primer set (supplied with the TwistDx RPA kit) does not work at this temperature, or it could be due to an experimental error. No amplification appears to occur at 4 °C, both in the sample and positive control.
In the previous experiment (Fig. 3), the incubator was opened every five minutes to take one sample out. This could have influenced the amplification process as the temperature was lowered every time the PCR machine was opened. Next, to monitor the amplification kinetics more accurately and more frequently, real-time RPA (SYBR Green dye) assays were performed at 30, 37 and 42 °C. Unfortunately, it was not possible to perform RPA at 20 °C due to time restrictions.
With real-time RPA, we confirmed the previous results that amplification can be performed at 30, 37 and 42 °C (Fig. 4). The rate of the reaction is higher at 37 °C than at 42 °C. While the reaction performed at 30 °C at some point reaches a reaction rate that is similar to the reaction performed at 37 °C, the reaction started significantly slower at 30 °C. Thus, for a reaction of 5 minutes, it is important to incubate at the optimal temperature (37 °C) to obtain a high amplification yield. In contrast, for a reaction of 10 minutes, the temperature is a less important factor as the amplification yield was similar for all temperatures at this time point. Thus, it is important to consider the tradeoff between a high amplification yield, a short reaction time and a low temperature when RPA is implemented in a point-of-care diagnostic test.
Fig. 4 Real-time RPA with S. cerevisiae primer set and genome. Forward primer (SC1_F_p1) and reverse primer (SC1_R_Nt.BstNBI_EAD2+3A) targeted S. cerevisiae BY4741 genome. Graphs were normalized by subtracting the fluorescence value at time point zero from all measurements for each temperature. Replicates: n = 3 for 30 °C and 37 °C, n = 2 for 42 °C.
RPA sensitivity and template concentrations
To test the sensitivity of RPA, RPA reactions were performed with different template concentrations. The concentration of the template stock was approximated using Nanodrop. RPA reactions with a template concentration ranging from 108 to 10-2copies were analyzed with real-time RPA using the SYBR Green dye and gel electrophoresis. The high copy numbers were included to determine how a template overload may affect RPA.
Both the real-time RPA results and gel electrophoresis results showed inconsistencies (Fig. S2). For instance, some template concentrations clearly suggested amplification of the target sequence, while other samples with higher template concentrations showed less or no amplification. The gel electrophoresis results also indicated the occurrence of off-target amplification; bands were obtained at 100 bp and lower, a sequence length that is significantly shorter than the targeted sequence of 191 bp (Fig. S2c and d). Off-target amplification may have been caused by primer-primer interactions and/or primer secondary structures (as reinforced later in Fig. 10b). The off-target amplification may explain the inconsistent results that we obtained from these sensitivity experiments. The results from these experiments were therefore treated as inconclusive.
Fig. S2 RPA sensitivity analysis with different template concentrations. Forward primer SC1_F_p1 and reverse primer SC1_R_Nt.AlwI_EAD2+3A) were used with template RPA_temp_SC. a) Real time RPA. From 1 till 8: 4.23*106 copies decreasing with 101 until 4.23*10-1 for 8; 9 is the negative control (no template). b) Real time RPA. From 1 till 4: 1106 copies to 1.106 copies in 4. No negative control. c) RPA analyzed with gel electrophoresis. From left to right: 1.11*108 copies to 1.11*11 copies represented by the final lane. d) RPA analyzed with gel electrophoresis. From left to right: 1106 copies to 0.0106 copies in the final lane.
Fig. S2 RPA sensitivity analysis with different template concentrations. Forward primer SC1_F_p1 and reverse primer SC1_R_Nt.AlwI_EAD2+3A) were used with template RPA_temp_SC. a) Real time RPA. From 1 till 8: 4.23*106 copies decreasing with 101 until 4.23*10-1 for 8; 9 is the negative control (no template). b) Real time RPA. From 1 till 4: 1106 copies to 1.106 copies in 4. No negative control. c) RPA analyzed with gel electrophoresis. From left to right: 1.11*108 copies to 1.11*11 copies represented by the final lane. d) RPA analyzed with gel electrophoresis. From left to right: 1106 copies to 0.0106 copies in the final lane.
RPA on various target organisms
To test the modularity of our system, we performed RPA reactions with small synthetic templates derived from different microorganisms and corresponding primer sets. Template and primer sequences can be found in Table 2 and 3 on the Parts page. Target microorganisms included S. cerevisiae, Bacillus subtilis, Plasmodium falciparum (malaria), Mycobacterium sp. and influenza virus A. Based on previous signals of off-target amplification (Fig. S2 and Fig. 10b), 3.5% DMSO was included in the RPA mix to enhance the selectivity of the amplification reaction by reducing primer-primer interactions and primer secondary structures.
Based on real-time RPA results, it can be surmised that the primer sets designed for S. cerevisiae and the influenza A virus allow for amplification with a low false-positive signal (Fig. S3). In contrast, the results for Bacillus subtilis, Plasmodium falciparum and Mycobacterium sp. showed a high false-positive signal. Interestingly, the real-time RPA graph of Plasmodium falciparum shows a false-positive signal for the negative sample but no signal increase for the positive sample. This may be the result of an experimental error, such as no addition of the SYBR Green dye.
Fig. S3 Real time RPA with primer set and synthetic template for a) S. cerevisiae (primers SC1_F_p1 and SC1_R_Nt.BstNBI_EAD2+3A), b) Influenza A virus (primers IA1_F_p1 and IA1_R_NtBstNBI_EAD2+3'A), c) Bacillus subtilis (primers BS1_F_p1 and BS1_R_Nt.BstNBI_EAD2+3A), d) Plasmodium falciparum (primers PF1_F_p1 and PF1_R_Nt.BstNBI_EAD2+3A) and e) Mycobacterium bovis (primers MB1_F_p1 and MB1_R_Nt.BstNBI_EAD2+3A). Primers targeted synthetic template sequences (Table 2 on Parts page). 3.5% DMSO was included in the RPA mix.
Fig. S3 Real time RPA with primer set and synthetic template for a) S. cerevisiae (primers SC1_F_p1 and SC1_R_Nt.BstNBI_EAD2+3A), b) Influenza A virus (primers IA1_F_p1 and IA1_R_NtBstNBI_EAD2+3'A), c) Bacillus subtilis (primers BS1_F_p1 and BS1_R_Nt.BstNBI_EAD2+3A), d) Plasmodium falciparum (primers PF1_F_p1 and PF1_R_Nt.BstNBI_EAD2+3A) and e) Mycobacterium bovis (primers MB1_F_p1 and MB1_R_Nt.BstNBI_EAD2+3A). Primers targeted synthetic template sequences (Table 2 on Parts page). 3.5% DMSO was included in the RPA mix.
RPA scale down
Decreasing the RPA reaction volume would also allow us to lower the cost of our next experiments in the lab and our final diagnostic test, thus making it more affordable for customers. Therefore, the RPA reactions were scaled down from 50 µL to 20 µL and 10 µL to test whether this would still lead to sufficient amplicon generation. The amplicons were analyzed by gel electrophoresis and Nanodrop. The amplification products for 10 and 20 µL reaction volume showed similar results as for 50 µL reaction volume (data not shown). These results suggested that RPA can be scaled down by 2.5-fold and even 5-fold. We continue our experiments with a reaction volume of 20 µL, since this method is less prone to pipetting errors than with a reaction volume of 10 µL.
LSDA
Confirmation of GQ DNAzyme generation with LSDA with different nickases
Text and figure adapted from our preprint.
Fig. 5 GQ DNAzyme production using SDLA reaction. SDLA was performed using different nickases. Here we identified Nt.BstNBI as the most potent GQ-producer. Synthetic (pure) oligonucleotide containing the targeted S. cerevisiae genome, nickase site, and EAD2+3’A DNAzyme sequences was used as template for the SDLA reaction.
In our sequence detection scheme, strand-displacement linear amplification (SDLA) acts as a connecting reaction to bridge initial signal generation by RPA and subsequent signal reporting by TMB oxidation. SDLA achieved this by producing GQ DNAzymes using purified RPA amplicons as templates. Here, we confirmed the effectivity of SDLA in guanine-quadruplex (GQ) sequence generation by using a synthetic (pure) template that resembles the RPA amplicon. The effectivity of Nt.BstNBI, Nt.Alwi, and Nt.BsmAI in mediating SDLA reactions was investigated. Two blank controls, which resembled the buffer conditions used in the three SDLA reactions, were included as standard of comparison. In the current reaction mix, we identified Nt.BstNBI as the most potent nickase for the production of GQ DNAzyme in an SDLA reaction (Fig. 5). This result confirmed the effectivity of the SDLA reaction in producing single-stranded GQ DNAzyme as well as its role as a bridge between RPA and TMB oxidation.
Scale down LSDA reaction
The most expensive part of the kit was initially estimated to be the commercially available nickase, which enables amplification of the GQ sequence in the LSDA reaction. To keep the cost of our diagnostic test at a minimum, it is essential that we reduce the amount of nickase in the reaction. Therefore, we performed two experiments to test if (1) the concentration of nickase can be reduced and (2) the reaction volume can be scaled down. The concentration of nickase in the LSDA reaction could be reduced from 1.28 U/µL to at least 0.16 U/µL (8-fold) (Fig. S4), and the reaction volume can be scaled down from 25 µL to 5 µL (5-fold) (Fig. S5). This means that the amount of nickase and thus the cost for the nickase in the test kit could be reduced 40-fold.
Fig. S4 LSDA reactions with different concentrations of nickase (Nt.BstNBI). The nickase concentration was varied to 0.16, 0.32 and 0.64 U/µL. The LSDA reaction with 0.16 U/µL nickase could still give significant color change.
Fig. S5 LSDA reactions with different reaction volumes. The reaction volume was varied to 5, 10 and 25 µL. Nt.BstNBI was used for all measurements. The LSDA reaction with 5 µL reaction volume could still give significant color change.
Oxidation
Confirmation of GQ-catalyzed TMB oxidation
Text and figure adapted from our preprint.
Guanine-quadruplex (GQ) sequences were known to have peroxidase-like activity when associated with hemin3. This allowed GQ-hemin complexes to be used as a catalyst for peroxidation reactions as an alternative to horseradish peroxidases3,4. Due to their enzyme-like ability, GQ-hemin complexes are often referred to as GQ DNAzymes. Just like horseradish peroxidases, GQ-hemin complexes can catalyze the color-changing oxidation reaction of 3, 3’, 5, 5’-tetramethylbenzedine (TMB) in the presence of peroxide. However, different GQ sequences may harbor different enzymatic capabilities. Our preliminary analysis identified EAD2+3’A as the most potent GQ DNAzyme (Fig. 6). EAD2+3’A was thus selected as the final signal reporter.
Fig. 6 TMB oxidation rate of different GQ DNAzyme sequences. Here, the catalytic activities of EAD2+3’A (BBa_K3343000), BBa_K1614007, and BBa_K1614007+3’A (BBa_K3343000) GQ DNAzymes in pH 6.0 phosphate buffer were compared. The DNAzyme sequences are shown in Table 1 on the Parts page. The concentration of DNAzyme was kept at 1 µM in all measurements. EAD2+3’A was identified as the most potent DNAzyme in catalysing TMB oxidation.
EAD2+3'A was added to the Registry as a New Part (BBa_K3343000). BBa_K3343001 was added to the Registry as an Improved Part .
Fig. 7 TMB oxidation by EAD2+3’A DNAzyme. Kinetic parameters of the DNAzyme were measured based on its ability to catalyze TMB oxidation reaction at different TMB concentrations. In the figure, data points represent initial oxidation rate measured from each experimental replicate while the curve represents model-predicted kinetics at each given TMB concentration. Rate kinetics of EAD2+3’A were assumed to follow Michaelis-Menten kinetics. The initial rate of each individual replicates was calculated based on differences in oxidized TMB absorbance value at 650 nm for the first minute of the observation. The resulting model could sufficiently describe each individual replicate measurement (Fig. S6).
Characterization of EAD2+3’A catalytic activity was performed based on the DNAzyme’s ability to oxidize TMB in the presence of an excess amount of peroxide and different concentrations of TMB. All reactions were performed in a pH 6.0 phosphate buffer. The rate of TMB oxidation was then monitored based on oxidized TMB’s absorbance at 650 nm wavelength (molar absorption coefficient is 39,000 M-1 cm-1)5. Kinetic parameters of the DNAzyme were described using the Michaelis-Menten equation (Eq. 1), with kcat and Km representing the molar maximum of TMB conversion rate and TMB affinity of EAD2+3’A, respectively. The values of kcat and Km were calculated using multiple regression analysis by simultaneously minimizing the error of all measurement-simulation pairs. The approximated kinetic parameters of EAD2+3’A against TMB are shown in Fig. 7, Fig. S6, and Table 1. We used these parameters in our (semi-)empirical model to simulate the underlying reaction kinetics of our detection technology.
Fig. 7 TMB oxidation by EAD2+3’A DNAzyme. Kinetic parameters of the DNAzyme were measured based on its ability to catalyze TMB oxidation reaction at different TMB concentrations. In the figure, data points represent initial oxidation rate measured from each experimental replicate while the curve represents model-predicted kinetics at each given TMB concentration. Rate kinetics of EAD2+3’A were assumed to follow Michaelis-Menten kinetics. The initial rate of each individual replicates was calculated based on differences in oxidized TMB absorbance value at 650 nm for the first minute of the observation. The resulting model could sufficiently describe each individual replicate measurement (Fig. S6).
Fig. S6 Predicted TMB oxidation kinetics by EAD2+3’A could sufficiently describe reaction rates of each individual reactions. The values above each figures represented TMB concentration used in each observation. Observations obtained from each experimental replicates were represented as data points. Reaction rate between each two observed time points were assumed to be constant and resulted to a linear increase in A650 value. Lines shown in the figure represent the simulated increase in A650 value over the observation period at a given TMB concentration, as predicted by the fitted kinetic model (Fig. 7), and by using the average of each measurement replicates at time point zero as starting point of the simulation.
References
- TwistDx™. TwistAmp DNA Amplification Kits Combined Instruction Manual. 2018. Retrieved from https://www.twistdx.co.uk/docs/default-source/RPA-assay-design/ta01cmanual-combined-manual_revo_v1-3b.pdf?sfvrsn=14
- Crannell, Z. A., Rohrman, B. & Richards-Kortum, R. Equipment-free incubation of recombinase polymerase amplification reactions using body heat. PLoS One 9, e112146 (2014).
- Li, W. et al. Insight into G-quadruplex-hemin DNAzyme/RNAzyme: Adjacent adenine as the intramolecular species for remarkable enhancement of enzymatic activity. Nucleic Acids Res. 44, 7373–7384 (2016).
- Yang, X. et al. Characterization of G-quadruplex/hemin peroxidase: Substrate specificity and inactivation kinetics. Chem. - A Eur. J. 17, 14475–14484 (2011).
- Liu, Y., Zhu, G., Yang, J., Yuan, A. & Shen, X. Peroxidase-like catalytic activity of Ag3PO4 nanocrystals prepared by a colloidal route. PLoS One 9, e109158 (2014).
Here we describe how the three main reactions of our detection method (RPA, LSDA and DNAzyme-catalyzed TMB oxidation) were coupled. The method’s feasibility was evaluated by passing the product of each preceding reactions to the next in a serial manner.
Text and figures adapted from our preprint.
Dithiothreitol (DTT) inhibits GQ oxidation but is inactivated at low pH
After coupling the reactions for the first time, a positive colorimetric signal could not be obtained using our original method and formulation (Fig. 8a and b). Interestingly, the signal could be obtained when RPA product was purified before being used in the subsequent LSDA reaction (Fig. S7). Thus, we suspected RPA mixture to contain a chemical(s) that may have inhibited LSDA and/or DNAzyme-catalyzed TMB oxidation.
The RPA mixture from TwistDx Basic Kit contained dithiothreitol (DTT), a strong reducing agent that is commonly mixed into protein formulations. The presence of DTT is known to improve protein stability by keeping the proteins’ monothiol groups in its reduced state1. Knowing its role as a reducing agent, we hypothesized that DTT carryover in TMB oxidation likely prevented the accumulation of oxidized TMB. Indeed, TMB oxidation with purified EAD2+3’A DNAzyme could not produce color change in the presence of DTT (Fig. S8). Therefore, removal or inactivation of DTT is essential for our DNA detection scheme to work.
The reducing power of DTT is known to be limited to more basic conditions2. Here, this characteristic of DTT was exploited as an alternative strategy to inactivate DTT during the DNAzyme-mediated TMB oxidation. Our original TMB oxidation formulation was performed using a pH 6.0 phosphate buffer. At this pH, we found that DTT hampers TMB oxidation (Fig. S8). Thus, we explored the possibility of DTT inactivation by performing DNAzyme-mediated TMB oxidation under different buffer conditions (Fig. 8c, d and e). Based on this comparison, phosphate-citrate buffer of pH 4.2 was identified to be the most suitable for TMB oxidation in the presence of DTT.
Fig. 8 Effect of DTT on TMB oxidation reaction. a and b) RPA mixture component prevented color generation in TMB oxidation. a) TMB oxidation of LSDA-treated unpurified RPA products could not generate observable color change. The same was also observable from the negative controls (NC), which were performed in the absence of RPA target sequence. b) in a parallel measurement, color generation could be observed when RPA mixture was not present in the final TMB oxidation mixture. Background color generation from the blank control was also significantly higher than that observed from the test variables, suggesting the presence of a color generation-inhibiting component in the RPA mixture. c, d and e) Acidic pH eliminates DTT reducing power. Here, we investigated the suitability of acidic c) phosphate-citrate buffer and d) phosphate-citrate buffer for GQ DNAzyme-catalyzed TMB oxidation reactions by comparing its performances to e) the original pH 6.0 phosphate buffer. In general, lower pH seemed to be favourable for signal production.
Fig. 8 Effect of DTT on TMB oxidation reaction. a and b) RPA mixture component prevented color generation in TMB oxidation. a) TMB oxidation of LSDA-treated unpurified RPA products could not generate observable color change. The same was also observable from the negative controls (NC), which were performed in the absence of RPA target sequence. b) in a parallel measurement, color generation could be observed when RPA mixture was not present in the final TMB oxidation mixture. Background color generation from the blank control was also significantly higher than that observed from the test variables, suggesting the presence of a color generation-inhibiting component in the RPA mixture. c, d and e) Acidic pH eliminates DTT reducing power. Here, we investigated the suitability of acidic c) phosphate-citrate buffer and d) phosphate-citrate buffer for GQ DNAzyme-catalyzed TMB oxidation reactions by comparing its performances to e) the original pH 6.0 phosphate buffer. In general, lower pH seemed to be favourable for signal production.
Fig. S7 TMB oxidation could be performed using LSDA product from purified RPA product. TMB oxidation progressed at a significantly higher rate in the presence of LSDA product compared to when the reaction was spiked with pure 4 µM EAD2+3’A DNAzyme. This suggested LSDA’s effectivity in producing EAD2+3’A to more than 4 µM within the 40-minutes incubation time. Moreover, this suggested the importance of RPA mixture removal in order to obtain TMB oxidation signal.
Fig. S8 DTT inhibits TMB oxidation reaction by EAD2+3’A. Absorbance signal at 650 nm of each spiked samples was not distinguishable from the background noise signal in the presence of 0.1 mM DTT.
Coupling RPA, LSDA, and TMB oxidation
Our attempt to serially couple RPA, LSDA, and TMB oxidation reaction in a pH 4.2 phosphate-citrate buffer did not produce a significant color signal. Although the low-pH buffer reduced DTT activity, the oxidation reaction was still inhibited by the presence of unpurified RPA mixture. Fig. 9a and b showed how the increasing amount of RPA mixture reduced the color generation rate during the oxidation reaction. The pellet mixture was found to completely inhibit color generation when diluted by less than 200-fold.
To further attenuate the oxidation-inhibiting effect of the RPA mixture, we decided to reduce the amount of RPA and LSDA product used in each of its subsequent reactions in order to obtain a final RPA mixture dilution rate of at least 625-fold. Additionally, we decided to further increase the acidity of phosphate-citrate buffer used in the TMB oxidation reaction to a final pH of 3.8 to anticipate for potential pH increase when the solution was mixed with LSDA product, which had a slightly basic pH. Here we explored the effectivity of different dilution schemes from RPA to SDA and from SDA to TMB oxidation. Based on this analysis, a dilution of 10-fold during transfer from RPA to SDA and a subsequent dilution of 80-fold during transfer from SDA to TMB oxidation was found to be the most favourable dilution scheme (Fig. 9c, d, and e); the change in absorbance was similar to the positive control that did not contain any RPA or SDA reagents (no DTT).
Serial coupling of RPA, LSDA, and TMB oxidation was achieved by using the new dilution scheme and pH adjustment (Fig. 9c and d). The resulting color signal was significantly higher compared to the blank controls. With the new dilution method and low-pH buffer, the TMB oxidation-inhibiting effects of compounds in the RPA and/or SDA reactions were diminished. Subsequently, this allowed successful coupling of RPA, SDA and TMB oxidation in a serial manner.
Fig. 9 Dilution of RPA mixture is required for signal generation in TMB oxidation reaction. a and b) increasing the amounts of RPA mixture (not incubated, no template) was found to significantly reduce signal generation during the TMB oxidation reaction. In these reactions, 1 µM EAD2+3’A DNAzyme was used to catalyze the reaction. c and d) during TMB oxidation, a signal generation rate comparable to that achieved from 1 µM EAD2+3’A DNAzyme could be obtained by reducing the amount of unpurified RPA product that is passed on to the LSDA reaction to one-fourth of its amount in the original scheme. Within the observed RPA product concentration range, the initial TMB oxidation rate appeared to be correlated to the amount of unpurified RPA product used in LSDA. e) serial coupling of RPA, LSDA, and TMB oxidation using the updated dilution scheme and buffer could produce observable color change.
Fig. 9 Dilution of RPA mixture is required for signal generation in TMB oxidation reaction. a and b) increasing the amounts of RPA mixture (not incubated, no template) was found to significantly reduce signal generation during the TMB oxidation reaction. In these reactions, 1 µM EAD2+3’A DNAzyme was used to catalyze the reaction. c and d) during TMB oxidation, a signal generation rate comparable to that achieved from 1 µM EAD2+3’A DNAzyme could be obtained by reducing the amount of unpurified RPA product that is passed on to the LSDA reaction to one-fourth of its amount in the original scheme. Within the observed RPA product concentration range, the initial TMB oxidation rate appeared to be correlated to the amount of unpurified RPA product used in LSDA. e) serial coupling of RPA, LSDA, and TMB oxidation using the updated dilution scheme and buffer could produce observable color change.
DMSO reduced primer dimer formation during RPA and improved method accuracy
Our serial RPA-LSDA-TMB oxidation method could also generate color signal even when the RPA reaction was performed in the absence of the targeted sequence (Fig. 10a). The false-positive signal was not observable when RPA mixture was not incubated. Similarly, the signal could not be generated when (pure) RPA primers were used as templates for the LSDA reaction (Fig. S9). This pinpointed the origin for the false signal to an RPA product. We suspected RPA product(s) of primer complexes to somehow generate a nickase site and a GQ sequence at the correct order, thus allowing LSDA to generate GQ DNAzyme from this side-product.
Real-time RPA analysis using SYBR Green dye captured the amplification of an unknown product in the absence of target template (Fig. 10b). This confirmed that our RPA primers alone could generate an amplicon. Interestingly, we also found that the non-tailed forward primer, and not the tailed reverse primer, could also generate a background amplicon on its own, even though the product may differ from that produced in the presence of both the forward and reverse primers.
As an attempt to reduce non-specific amplicon generation by primer complexes, dimethyl sulfoxide (DMSO) was added to the RPA reactions. DMSO is a polar aprotic solvent. Its ability to dissolve both polar and non-polar compounds allows it to disrupt secondary structure formation of DNA strands. By weakening the hydrogen bonds between primers, the presence of DMSO may prevent primer (self-) annealing3,4. Here we exploit this capability of DMSO to weaken the non-specific primer-primer binding and complex formation (Fig. 10c). In this sense, DMSO acted much like a filter – eliminating background noise while allowing significant levels of true-positive signal to be further amplified. Here we identified 2.5% DMSO to be the most suitable for our RPA primer pair.
The inclusion of 2.5% DMSO in the RPA mixture was found to be effective in reducing the occurrence of false-positive signal from our sequence detection method. Significant attenuation of the true-positive signal was not observable, confirming the strategy’s effectivity to selectively filter out most false-positive amplicons (Fig. 10d). False-positive cases could nevertheless still be observable. We suspected this was due to the residual production of primer-associated amplicons. Although the current RPA condition could prevent false-positive signal production in RT-RPA, purified product from this reaction could still give a smear at approximately 100 bp region in agarose gel electrophoresis (Fig. S10). Nevertheless, the current method showed 33.3% false-positive and 33.3% false-negative rates based on production of absorbance signals higher than 0.15 (a.u.) within the given test period. This suggested the need for further improvements and technique optimization.
Interestingly, the maximum readout of true-positive signals was significantly lower compared to that observed in our previous experiments. This can be due to the interaction between DMSO and components of the LSDA and/or TMB oxidation reactions. Further analysis is required to pinpoint the cause of this signal attenuation and to further optimize our sequence detection technique.
Fig. 10 DMSO reduced primer dimer formation during RPA and improved method accuracy. a and b) RPA amplification product in the absence of target template can produce a positive signal. a) The average endpoint colorimetry readout of negative tests was approximately was approximately twice as high as was observed in the blank control. In this figure, error bars represented standard deviation of three (positive and negative tests) or two (blank and EAD2+3’A control) replicates. b) RT-RPA analysis confirmed the occurrence of background amplicon production in the absence of target template. Production of this background amplicon appeared to be dependent on the forward and not the reverse (tailed) primer. c) Addition of DMSO into RPA mixture could eliminate background amplicon generation. Though effective, high concentrations of DMSO (5% v/v or higher) may also attenuate amplicon generation in the absence of target templates. Here we identified 2.5% DMSO to be the most suitable to distinguish the presence of template during the RPA reaction. d) 2.5% DMSO could significantly reduce false-positive signal from the overall sequence detection method. Nevertheless, significant production of absorbance signal could still be observed 33.3% of the time. This may be due to residual production of false-positive-associated amplicons (Fig. S10). Average signals generated from positive- and negative-tests are shown for comparison with the EAD2+3’A and the blank control. Individual measurements are shown in Fig. S11.
Fig. 10 DMSO reduced primer dimer formation during RPA and improved method accuracy. a and b) RPA amplification product in the absence of target template can produce a positive signal. a) The average endpoint colorimetry readout of negative tests was approximately was approximately twice as high as was observed in the blank control. In this figure, error bars represented standard deviation of three (positive and negative tests) or two (blank and EAD2+3’A control) replicates. b) RT-RPA analysis confirmed the occurrence of background amplicon production in the absence of target template. Production of this background amplicon appeared to be dependent on the forward and not the reverse (tailed) primer. c) Addition of DMSO into RPA mixture could eliminate background amplicon generation. Though effective, high concentrations of DMSO (5% v/v or higher) may also attenuate amplicon generation in the absence of target templates. Here we identified 2.5% DMSO to be the most suitable to distinguish the presence of template during the RPA reaction. d) 2.5% DMSO could significantly reduce false-positive signal from the overall sequence detection method. Nevertheless, significant production of absorbance signal could still be observed 33.3% of the time. This may be due to residual production of false-positive-associated amplicons (Fig. S10). Average signals generated from positive- and negative-tests are shown for comparison with the EAD2+3’A and the blank control. Individual measurements are shown in Fig. S11.
Fig. S9 False-positive signal originated from RPA side-product in the absence of target template DNA. Without incubation, the same no-template RPA mix could not produce similarly high signal during TMB oxidation (triangles). The same is also true when either primer sets or RPA mixture was missing.
Fig. S10 Faint band at approximately 150 bp region could be observed from purified RPA product in the absence of target template and in the presence of 2.5% DMSO. Although this band could only become visible at a higher resolution, this suggested residual production of false-positive-associated amplicons that may result to an increase in our method’s false-positive rate.
Fig. S11 2.5% DMSO could significantly reduce false-positive signal from the overall sequence detection method. a) Positive RPA reactions (with target sequence) and b) negative RPA reactions (without target sequence) were performed in presence of 2.5% DMSO and analyzed with LSDA and GQ-catalyzed oxidation. Average signals generated from each positive- and negative-tests are shown in Fig. 4d for comparison with the EAD2+3’A and the blank control.
References
- Cleland, W. W. Dithiothreitol, a New Protective Reagent for SH Groups. Biochemistry 3, 480–2 (1964).
- Lukesh, J. C., Palte, M. J. & Raines, R. T. A potent, versatile disulfide-reducing agent from aspartic acid. J. Am. Chem. Soc. 134, 4057–4059 (2012).
- Hardjasa, A., Ling, M., Ma, K. & Yu, H. Investigating the Effects of DMSO on PCR Fidelity Using a Restriction Digest-Based Method. J. Exp. Microbiol. Immunol. 14, 161–164 (2010).
- Nakano, S. ichi & Sugimoto, N. The structural stability and catalytic activity of DNA and RNA oligonucleotides in the presence of organic solvents. Biophysical Reviews 8, 11–23 (2016).